3
\$\begingroup\$

At first I thought tone mapping was a thing for HDR images only used to "display" the dynamic range which our monitors can't show. Turns out that RAW images tend to have a high dynamic range too. So they are actually HDR images too. And now I'm thinking:

  1. What do I actually see when I look at the preview from my Camera or LR? Where is the extra range? is it clipped? or already tone mapped?

  2. What happens when I slide the shadows slider up or the highlights down? what is LR doing to "bring back the highlights"?

  3. (bonus Q) are the occasional Halos on RAW caused by the tone mapping?

EDIT:

I did some more reading and this may be seam unrelated to my question, but the answer that I was actually looking for was: gamma correction is a method/form of tone mapping. That made so many things clearer for me.

Improve this question
\$\endgroup\$
4
  • \$\begingroup\$ RAW images don't have any dynamic range at all, high or low. What they have is a certain bit depth. With 8 bits you have 256 values between 0 (full black) and 1 (full white). With 14 bits you have 16384 values. It's the sensor (combined with the rest of the optical path - lenses, filters, etc.) that is characterized by its ability to capture more or less range of light, and the camera circuits and firmware that are responsible for mapping that range into the bits available as well as it can. \$\endgroup\$
    – twalberg
    May 15, 2019 at 14:15
  • 4
    \$\begingroup\$ Yes, a certain bit depth. but that bit depth translates to a dynamic range (either linearly or non-linearity). Saying RAW images don't have a dynamic range may be correct but only in the literal manner. \$\endgroup\$
    – Fanckush
    May 15, 2019 at 15:00
  • 2
    \$\begingroup\$ Thanks! These all seam like great pointers! I think it fits to write an answer with all these links. \$\endgroup\$
    – Fanckush
    May 15, 2019 at 17:25
  • 1
    \$\begingroup\$ @twalberg RAW images DO have a dynamic range. Any image, or signal, has a dynamic range. And yes, RAW images have a greater dynamic range than jpg images. en.wikipedia.org/wiki/Dynamic_range \$\endgroup\$
    – roetnig
    May 16, 2019 at 10:03

2 Answers 2

Reset to default
5
\$\begingroup\$

This question is probably too broad for the format here and might be closed as such. But it becomes a little unwieldy to place links to even a few of the many existing questions/answers we have here that touch on the various questions you raise in comments. This isn't really so much a complete answer as it is an attempt to point you in a direction where these subjects have already been addressed here.

Is tone mapping automatically applied to RAW images?

A lot of processing is applied to the raw image data collected by the camera's sensor and digitized by the camera's processor before there is what we consider a "viewable image." This processing include what you refer to as "tone mapping."

What do I actually see when I look at the preview from my Camera or LR? Where is the extra range? is it clipped? or already tone mapped?

For more about what a raw file is and what a raw file isn't (Hint: the image you see on your camera's LCD screen or on your computer is not an unprocessed raw image), please see:

RAW files store 3 colors per pixel, or only one?
What does an unprocessed RAW file look like?
Is the Preview file always the photo taken by the camera?
If I save as RAW+JPG, which of the two is shown on the screen of a Canon 600D?
Why are my RAW images already in colour if debayering is not done yet?

What happens when I slide the shadows slider up or the highlights down? what is LR doing to "bring back the highlights"?

What you are looking at on your screen is only one of countless possible interpretations of the raw image data. This preview is based on the instructions that have been given to Lightroom as to how the raw image data should be interpreted. When you move a slider, you change the set of instructions with regard to whatever the slider you moved controls. Lightroom then either:

  • Estimates the effect this change would have on reprocessing the raw data with the modified instruction set and renders this change on your screen

or

  • Reprocesses the raw image data with the new instruction set and displays the result on your screen.

Which one LR does is determined by your user settings regarding "quality" and "speed' in the 'User preferences' section covering rendering of the preview you see on your screen when working with raw files. In either case, when you actually tell LR to export the file, it will use the instruction set you have designated to go back and reprocess the raw image data to produce a jpeg, png, tiff, etc. for export.

For more about what happens when you move a slider while processing a raw image file in Lightroom or other raw processing applications, please see:

Camera dynamic range and Lightroom exposure slide
Why can software correct white balance more accurately for RAW files than it can with JPEGs?

Are the occasional Halos on RAW caused by the tone mapping?

In short: Yes, halos can be a result of global tone-mapping, whether done via an "HDR" application or by aggressively tone-mapping a single raw file.

enter image description here

Regarding how global tone-mapping can lead to halos in boundary areas between very light and very dark areas, please see:

What is the difference between Digital Blending & HDR technique

Improve this answer
\$\endgroup\$
3
  • \$\begingroup\$ ".. it will use the instructions.." a silly question but I really need to know if global tone mapping is part of these instructions? if so does that mean that other RAW editing apps can choose other "HDR to 8-bit" methods other besides tone mapping? \$\endgroup\$
    – Fanckush
    May 15, 2019 at 17:47
  • \$\begingroup\$ Tone mapping was around long before digital photography, much less 32-bit floating point light maps that many people think are the only way to do HDR. The darkroom techniques that Ansel Adams raised to a high art form in the 1930s was "tone mapping' (though not "global"). Any method that transforms information in a way that the things that are darker than other things in the information source (i.e. negative or raw image file) aren't always darker than the same things in the output image is a type of tone mapping. Gamma correction is very close, and some forms of sharpening are in fact. \$\endgroup\$
    – Michael C
    May 15, 2019 at 19:35
  • \$\begingroup\$ you must've heard this before, but if you write a book about all those core technicalities (gamma curves, tone mapping, what is RAW fundamentally, why 2 green on bayer filter etc.) that many photographs don't even think about, I'll pre order it!! \$\endgroup\$
    – Fanckush
    May 16, 2019 at 9:21
1
\$\begingroup\$

  1. The range is more like rounded off. With a 14-bit sensor and a 8-bit display, each range of 64 (2^6) sensor values becomes a single value in the displayed 0..255 range (values 0..63 become 0, 64..127 become 1, and so forth (it's actually a bit more complicated because before this a gamma correction is applied).

  2. LR has more data to work with than what it can show. If you want to lighten the shadows, you can for instance map the range 0..31 to 0, 32..63 to 1, 64..95 to 2, and 96..127 to 4. So the values that are displayed are still a continuous set... If you were to try this with an 8-bit JPEG, where one value of data is one value of display, your 0 would stay 0, but your 1 would be 4, your 2 would be 8, your 3 would be 12 and you would start to see bands. A symmetric effect exists at the other end of the range.

  3. No, this is possibly an artifact of the demosaicing alogtithm (or caused by sensor blooming).

Improve this answer
\$\endgroup\$
1
  • \$\begingroup\$ The 14-bit raw values are monochromatic. There is only one brightness value per photosite/sensel/"pixel well". The 8-bit values for a display are 8-bits per channel for each set of three subpixels, for a total of 24-bits per pixel. Though both sets of information are recorded using bits, as all digital data is, those bits do not represent the same thing. It's not just a case of reducing 14-bit values down to 8-bit values. It's a case of interpolating 8-bit values for three color channels using demosaicing/debayering by comparing the relative brightness values of adjoining sensels. \$\endgroup\$
    – Michael C
    Aug 23 at 22:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.